提交 · 6d8771b55c610082ca26ba6acf498a0915e1eb55 · 机器未来 / Paddle

05 3月, 2019 3 次提交
- B
  
  fix tanh typo test=develop (#16049) · da45fbda
  由 baojun 提交于 3月 05, 2019
  
  da45fbda
- W
  Make sequence_erase op support for input with multi-level LoD. (#15982) · 0f99d240
  由 whs 提交于 3月 05, 2019
```
test=develop
```
  0f99d240
- L
  MKLDNN: Add UT for conv_transpose_mkldnn op. (#16030) · 02c106c7
  由 lidanqing 提交于 3月 05, 2019
```
* MKLDNN: Add UT for conv_transpose_mkldnn op.
test=develop

* MKLDNN: Add fuse_bias check UT for conv_transpose_mkldnn op.
test=develop
```
  02c106c7
04 3月, 2019 13 次提交
- D
  
  fix statement. test=develop · 3eab9e4b
  由 dengkaipeng 提交于 3月 04, 2019
  
  3eab9e4b
- D
  
  fix API.spec. test=develop · e37f5ab5
  由 dengkaipeng 提交于 3月 04, 2019
  
  e37f5ab5
- D
  
  fix doc statement. test=develop · 54bbbfa7
  由 dengkaipeng 提交于 3月 02, 2019
  
  54bbbfa7
- D
  
  refine doc. test=develop · c1a69e3e
  由 dengkaipeng 提交于 3月 01, 2019
  
  c1a69e3e
- D
  
  fix format. test=develop · 65d375a0
  由 dengkaipeng 提交于 2月 27, 2019
  
  65d375a0
- D
  
  fix spectral_norm doc. test=develop · 82d51434
  由 dengkaipeng 提交于 2月 27, 2019
  
  82d51434
- D
  
  add doc and test_layers. test=develop · 2ea5843c
  由 dengkaipeng 提交于 2月 21, 2019
  
  2ea5843c
- D
  
  fix attr dim calc. test=develop · 037855f4
  由 dengkaipeng 提交于 2月 21, 2019
  
  037855f4
- D
  
  add grad kernel for spectral_norm. test=develop · 70dbd598
  由 dengkaipeng 提交于 2月 21, 2019
  
  70dbd598
- D
  
  add unittest for spectral_norm. test=develop · 72509ec3
  由 dengkaipeng 提交于 2月 21, 2019
  
  72509ec3
- D
  
  add spectral_norm forwarn kenel · 3bf1ae9b
  由 dengkaipeng 提交于 2月 20, 2019
  
  3bf1ae9b
- C
  Revert "Add Event for TensorCopy" (#16022) · e2da3a5b
  由 chengduo 提交于 3月 03, 2019
```
* Revert "Add Event for TensorCopy (#15953)"

This reverts commit 7235fd66.
test=develop

* fix CI
test=develop
```
  e2da3a5b
- B
  
  fix cpplint test=develop (#16028) · 9aaea38c
  由 baojun 提交于 3月 03, 2019
  
  9aaea38c
01 3月, 2019 2 次提交
- C
  Add Event for TensorCopy (#15953) · 7235fd66
  由 chengduo 提交于 3月 01, 2019
```
Add Event for TensorCopy 
```
  7235fd66
- T
  refine image_resize annotation (#15976) · 31d830de
  由 Tink_Y 提交于 3月 01, 2019
```
* fix image_resize annotation

test=develop

* fix some typo

* Update nn.py

* Update interpolate_op.cc

test=develop
```
  31d830de
28 2月, 2019 2 次提交
- Y
  Revert "Optimize while_op when is_test is true. (#15811)" (#15968) · 79892545
  由 Yiqun Liu 提交于 2月 28, 2019
```
test=develop
```
  79892545
- Y
  Fix error in CUDA kernel of beam_search. (#15957) · 87248281
  由 Yiqun Liu 提交于 2月 28, 2019
```
test=develop
```
  87248281
27 2月, 2019 9 次提交
- Y
  Optimize while_op when is_test is true. (#15811) · 613d9d07
  由 Yiqun Liu 提交于 2月 27, 2019
```
test=develop
```
  613d9d07
- X
  Optimize Quantize Op with primitive reuse. (#15929) · 1abddd8d
  由 xiaolil1 提交于 2月 27, 2019
```
test=develop
```
  1abddd8d
- L
  refine infershape of sequence_enumerate, hash and fuse_emb_seq_pool · 34404f9c
  由 luotao1 提交于 2月 27, 2019
```
test=develop
```
  34404f9c
- B
  
  Added adam op test=develop (#15710) · f285191f
  由 baojun 提交于 2月 26, 2019
  
  f285191f
- M
  Register sum operator (#15889) · 558f94cd
  由 mozga-intel 提交于 2月 27, 2019
```
test=develop
```
  558f94cd
- D
  polish cudnn related code and fix bug. (#15164) · 225c11a9
  由 dzhwinter 提交于 2月 27, 2019
```
* staged.

* polish code

* polish code. test=develop

* polish code. test=develop

* api change. test=develop

* fix default value. test=develop

* fix default value. test=develop
```
  225c11a9
- Y
  Rewrite is_empty op to avoid unnecessary data transform. (#15509) · 454f4f21
  由 Yiqun Liu 提交于 2月 27, 2019
```
* Rewrite is_empty op to avoid unnecessary data transform.
test=develop

* Add the implementation of InferShape and InferVarType for is_empty op.
test=develop

* Rewrite is_empty op to avoid directly inherit OperatorBase.
test=develop
```
  454f4f21
- X
  INT8 Pool kernel Key Creation Optimization. (#15883) · 6724be2b
  由 xiaolil1 提交于 2月 27, 2019
```
* Optimize key creation of INT8 pool kernel to improve the peformance of ResNet-50 and MobileNet, especially for latency.
test=develop

* Optimize key creation of pool fp32 grad.
test=develop
```
  6724be2b
- B
  
  added concat op test=develop · e4ab40a7
  由 baojun-nervana 提交于 2月 26, 2019
  
  e4ab40a7
26 2月, 2019 11 次提交

T
fix jitcodekey and refine test · 8bc63815
由 tensor-tang 提交于 2月 26, 2019
```
test=develop
```
8bc63815
T
add sgd jitcode and op test · 7044cfa7
由 tensor-tang 提交于 2月 25, 2019
```
test=develop
```
7044cfa7
T
add benchmark and mkl sgd implement · 8e041337
由 tensor-tang 提交于 2月 25, 2019
```
test=develop
```
8e041337

- MKL-DNN pooling updated to set_prim_desc · c63f6b20

由 Jacek Czaja 提交于 2月 04, 2019

- MKLDNN ops revisited

- disabled softmax modifications

- disabled elementwise_add

- reverted LRN modifications

- reverted SUM primitive

- Partial reviing of softmax

- Enable softmax

- Softmax changes

- LRN is back

- LRN partially disabled

- LRN is back

- LRN fix

- compilation fixes

- Sum fixed(hopefully)

- Enabling (partially) elementwise_add

- Fixes to elemenwise_add

- Lint fixes

quantize fix

- compilation fix

test=develop

Disabling pooling

- Disabled quantize op

test=develop

c63f6b20

Q

Fix bug in fake_quantize_op and add more unit testing (#15912) · 8e439ccf
由 qingqing01 提交于 2月 26, 2019

8e439ccf

loosly check in the InferShape of cross_entropy_op. (#15863) · f4846bf3

由 qingqing01 提交于 2月 26, 2019

* loosly check in cross_entropy_op when soft_label is True
* Add Runtime assertion in backward infer_shape check.
* Skip InferShape check when un-know the input dimensions

f4846bf3

Y
Optimize gelu operation with mkl erf. · 73967886
由 Yihua Xu 提交于 2月 26, 2019
```
test=develop
```
73967886
X
Optimize INT8 DeQuantize Op with primitive reuse. · 70759d18
由 xiaoli.liu@intel.com 提交于 2月 26, 2019
```
test=develop
```
70759d18

Optimize the CUDA implementation of sequence_expand op by reduce the times of... · f4634d76

由 Yiqun Liu 提交于 2月 26, 2019

Optimize the CUDA implementation of sequence_expand op by reduce the times of copying lod data from CPU to GPU. (#15493)

* Optimize the CUDA implementation of sequence_expand op by reduce the times of copying lod data from CPU to GPU.
test=develop

* Refine the op benchmark to support setting lod in config.
test=develop

f4634d76

This PR improve performance of prior_box op about 1.25x faster on CPU. (#15909) · 630c1e83

由 guomingz 提交于 2月 26, 2019

* This PR improve performance of prior_box op about 1.25x faster on CPU.

* Test Env:SKX 8180 with fake data on 28 threads(bs=1).
* The below table shows the ~25% improvement which generated by [eval_tp_fake_data.py](https://github.com/PaddlePaddle/Paddle/issues/15618#issuecomment-464613976).

| Type |Event | Calls |   Total     |  Min.    |   Max.      |  Ave.      |  Ratio.|
| ---------------- | ------------------ | ---- | ------- | -------- | -------- | ------------ | -------- |
| w/ optimization  | thread0::prior_box | 6000 | 921.201 | 0.110572 | 0.383402 | **0.153533** | 0.084585 |
| w/o optimization | thread0::prior_box | 6000 | 1151.85 | 0.102276 | 0.426702 | **0.191976** | 0.103337 |

test=develop

* Fix the style issue.

test=develop

630c1e83

Add alloc_continuous_space_op (#15900) · 7ca8553d

由 chengduo 提交于 2月 25, 2019

* add alloc_continuous_space_op
test=develop

* Polish code
test=develop

* follow comment
test=develop

7ca8553d

机器未来 / Paddle 与 Fork 源项目一致

机器未来 / Paddle
与 Fork 源项目一致